PUB - Publikationen an der Universität Bielefeld

This data publication contains all material used in Hötte, K., 2019, "Skill transferability and the adoption of new technology: A learning based explanation for patterns of diffusion". It is composed of (1) the simulation model are required inputs to reproduce the results, (2) simulated data presented in the article, (3) R-scripts that were used for the statistical analyses, (4) selected results and graphics that are partly used in the article and partly supplementary.

Please check for software updates (concerning the model and [R code](https://gitlab.ub.uni-bielefeld.de/khoette/rcode_eurace)) on gitlab. If you are only interested in the programming code, I recommend to check out gitlab first because this data publication consumes a lot of disk-space due to the large amount of simulated data (~16Gb).

If you have questions, do not hesitate to send me an email: kerstin.hoette[at]uni-bielefeld.de

### Background: Abstract of the underlying research paper Technological capabilities are decisive to make effectively use of new machinery and capital goods. Firms and employees accumulate these capabilities when working with specific machinery. Radical innovation differs by technology type and pre-existing capabilities may be imperfectly transferable across types. In this paper, I address the implications of cross-technology transferability of capabilities for firm-level technology adoption and macroeconomic directed technological change. I propose a microeconomically founded model of technological learning that is based on empirical and theoretical insights of the innovation literature. In a simulation study using the ABM Eurace@unibi-eco and applied to the context of green technology diffusion, it is shown that a high transferability of knowledge has ambiguous effects. It accelerates the diffusion process initially, but comes of the cost of technological stability and specialization. For firms, it is easy to adopt, but also easy to switch back to the conventional technology type. It is shown how different types of policies can be used to stabilize the diffusion process. The framework of analysis is used to derive a general characterization of technologies that may provide guidance for future empirical analyses. ### More detailed overview of the content: See also readme files in the subfolders. #### model: Simulation inputs The data provided should allow you to REPRODUCE the simulations, i.e. to produce your own simulation data that should exhibit the same patterns as those discussed in the paper. - model files: model xml files and c code Extensions compared to "standard eurace@unibi": (Here only main modifications highlighted) - Cons_Goods_UNIBI: Production routine and investment decision modified. - Government_GREQAM: Eco policy module added. - Inv_Goods_Vintage: Adaptive pricing mechanism and endogenous innovation added/ modified. - Labour_UNIBI: Households skill endowment and learning. - Statistical_Office_UNIBI: Documentation of indicator variables. - my_library_functions.c: Running order of vintages adjusted by using costs. - its: initial population #### experiment_directories_and_data: Simulated data (raw). This data allows you to perform STATISTICAL ANALYSES with the simulation output yourself. You may use these as input to the Rcode. Experiment folders contain simulation files and simulation output - baseline -> With intermediate techn. difficulty and distance - difficulty -> 3 discrete levels of chi^{dist} - distance -> 3 discrete levels of chi^{int} monte_carlo_exp: - both_learning_at_random_3_barr -> Monte Carlo analysis (MC) with fix barrier and randomly drawn learning parameters - both_learning_at_random_random_barr -> MC with random learning and random barrier at max 10 pct, serves as policy baseline - rand_learn_rand_pol_rand_barr10 -> Policy experiment In principle, you should be able to reproduce the simulated data (Note that the model has stochastic components, hence it will not be EXACTLY the same but sufficiently similar). #### rcode: This documentation makes the STATISTICAL METHODS used in the paper transparent. Sorry for the inefficient code. - rcode: R scripts used for statistical analysis of simulation output #### selected_results Output of analysed data and additional time series plots. Here, you find additional time series that are not presented in the main article, some descriptive statistics and the ouput of statistical tests and analyses, i.e. regression output and wilcoxon test results in txt format and plots that are used in the paper. These files can be reproduced by the R code. ### Acknowledgements The author gratefully acknowledges the achievements and provision of free statistical software maintained by the R programming community. This work uses a modified version of the Eurace@Unibi model, developed by Herbert Dawid, Simon Gemkow, Philipp Harting, Sander van der Hoog and Michael Neugart, as an extension of the research within the EU 6th Framework Project Eurace. Particular gratitude is owed to Herbert Dawid for uncomplicated and efficient support. Further, many thanks to the center for research data for support and the provision of excellent infrastructure. Particular thanks to Cord Wiljes.